Cover Detail
Show a cover letter.
GET /api/covers/51305/?format=api
{ "id": 51305, "url": "http://patches.dpdk.org/api/covers/51305/?format=api", "web_url": "http://patches.dpdk.org/project/dpdk/cover/20190319012010.16793-1-gage.eads@intel.com/", "project": { "id": 1, "url": "http://patches.dpdk.org/api/projects/1/?format=api", "name": "DPDK", "link_name": "dpdk", "list_id": "dev.dpdk.org", "list_email": "dev@dpdk.org", "web_url": "http://core.dpdk.org", "scm_url": "git://dpdk.org/dpdk", "webscm_url": "http://git.dpdk.org/dpdk", "list_archive_url": "https://inbox.dpdk.org/dev", "list_archive_url_format": "https://inbox.dpdk.org/dev/{}", "commit_url_format": "" }, "msgid": "<20190319012010.16793-1-gage.eads@intel.com>", "list_archive_url": "https://inbox.dpdk.org/dev/20190319012010.16793-1-gage.eads@intel.com", "date": "2019-03-19T01:20:04", "name": "[v8,0/6] Add lock-free ring and mempool handler", "submitter": { "id": 586, "url": "http://patches.dpdk.org/api/people/586/?format=api", "name": "Eads, Gage", "email": "gage.eads@intel.com" }, "mbox": "http://patches.dpdk.org/project/dpdk/cover/20190319012010.16793-1-gage.eads@intel.com/mbox/", "series": [ { "id": 3791, "url": "http://patches.dpdk.org/api/series/3791/?format=api", "web_url": "http://patches.dpdk.org/project/dpdk/list/?series=3791", "date": "2019-03-19T01:20:04", "name": "Add lock-free ring and mempool handler", "version": 8, "mbox": "http://patches.dpdk.org/series/3791/mbox/" } ], "comments": "http://patches.dpdk.org/api/covers/51305/comments/", "headers": { "Return-Path": "<dev-bounces@dpdk.org>", "X-Original-To": "patchwork@dpdk.org", "Delivered-To": "patchwork@dpdk.org", "Received": [ "from [92.243.14.124] (localhost [127.0.0.1])\n\tby dpdk.org (Postfix) with ESMTP id 326FB2B9E;\n\tTue, 19 Mar 2019 02:20:41 +0100 (CET)", "from mga02.intel.com (mga02.intel.com [134.134.136.20])\n\tby dpdk.org (Postfix) with ESMTP id AB09B152A\n\tfor <dev@dpdk.org>; Tue, 19 Mar 2019 02:20:39 +0100 (CET)", "from orsmga008.jf.intel.com ([10.7.209.65])\n\tby orsmga101.jf.intel.com with ESMTP/TLS/DHE-RSA-AES256-GCM-SHA384;\n\t18 Mar 2019 18:20:38 -0700", "from txasoft-yocto.an.intel.com ([10.123.72.192])\n\tby orsmga008.jf.intel.com with ESMTP; 18 Mar 2019 18:20:37 -0700" ], "X-Amp-Result": "SKIPPED(no attachment in message)", "X-Amp-File-Uploaded": "False", "X-ExtLoop1": "1", "X-IronPort-AV": "E=Sophos;i=\"5.58,495,1544515200\"; d=\"scan'208\";a=\"126564235\"", "From": "Gage Eads <gage.eads@intel.com>", "To": "dev@dpdk.org", "Cc": "olivier.matz@6wind.com, arybchenko@solarflare.com,\n\tbruce.richardson@intel.com, konstantin.ananyev@intel.com,\n\tstephen@networkplumber.org, jerinj@marvell.com, mczekaj@marvell.com, \n\tnd@arm.com, Ola.Liljedahl@arm.com, gage.eads@intel.com", "Date": "Mon, 18 Mar 2019 20:20:04 -0500", "Message-Id": "<20190319012010.16793-1-gage.eads@intel.com>", "X-Mailer": "git-send-email 2.13.6", "In-Reply-To": "<20190318213555.17345-1-gage.eads@intel.com>", "References": "<20190318213555.17345-1-gage.eads@intel.com>", "Subject": "[dpdk-dev] [PATCH v8 0/6] Add lock-free ring and mempool handler", "X-BeenThere": "dev@dpdk.org", "X-Mailman-Version": "2.1.15", "Precedence": "list", "List-Id": "DPDK patches and discussions <dev.dpdk.org>", "List-Unsubscribe": "<https://mails.dpdk.org/options/dev>,\n\t<mailto:dev-request@dpdk.org?subject=unsubscribe>", "List-Archive": "<http://mails.dpdk.org/archives/dev/>", "List-Post": "<mailto:dev@dpdk.org>", "List-Help": "<mailto:dev-request@dpdk.org?subject=help>", "List-Subscribe": "<https://mails.dpdk.org/listinfo/dev>,\n\t<mailto:dev-request@dpdk.org?subject=subscribe>", "Errors-To": "dev-bounces@dpdk.org", "Sender": "\"dev\" <dev-bounces@dpdk.org>" }, "content": "For some users, the rte ring's \"non-preemptive\" constraint is not acceptable;\nfor example, if the application uses a mixture of pinned high-priority threads\nand multiplexed low-priority threads that share a mempool.\n\nThis patchset introduces a lock-free ring and a mempool based on it. The\nlock-free algorithm relies on a double-pointer compare-and-swap, so for 64-bit\narchitectures it is currently limited to x86_64.\n\nThe ring uses more compare-and-swap atomic operations than the regular rte ring:\nWith no contention, an enqueue of n pointers uses (1 + n) CAS operations and a\ndequeue of n pointers uses 1. This algorithm has worse average-case performance\nthan the regular rte ring (particularly a highly-contended ring with large bulk\naccesses), however:\n- For applications with preemptible pthreads, the regular rte ring's worst-case\n performance (i.e. one thread being preempted in the update_tail() critical\n section) is much worse than the lock-free ring's.\n- Software caching can mitigate the average case performance for ring-based\n algorithms. For example, a lock-free ring based mempool (a likely use case\n for this ring) with per-thread caching.\n\nThe lock-free ring is enabled via a new flag, RING_F_LF. For ease-of-use,\nexisting ring enqueue/dequeue functions work with both standard and lock-free\nrings. This is also an experimental API, so RING_F_LF users must build with the\nALLOW_EXPERIMENTAL_API flag.\n\nThis patchset also adds lock-free versions of ring_autotest and\nring_perf_autotest, and a lock-free ring based mempool.\n\nThis patchset makes one API change; a deprecation notice was posted in a\nseparate commit[1].\n\nThis patchset depends on the 128-bit compare-and-set patch[2].\n\n[1] http://mails.dpdk.org/archives/dev/2019-February/124321.html\n[2] http://mails.dpdk.org/archives/dev/2019-March/125751.html\n\nv8:\n- Fixed two bugs in the generic implementation written as though the\n compare-and-swap would update the expected value on failure.\n\nv7:\n- Added ARM copyright to rte_ring_generic.h and rte_ring_c11_mem.h, since the\n lock-free algorithm is based on ARM's lfring (see v5 notes)\n- Rename __rte_ring_reload_tail() -> __rte_ring_lf_load_tail()\n- Remove the unused return value from __rte_ring_lf_load_tail()\n- Rename 'prev_tail' to 'next_tail' in the multi-producer lock-free enqueue\n\nv6:\n- Rebase patchset onto master (test/test/ -> app/test/)\n\nv5:\n - Incorporated lfring's enqueue and dequeue logic from\n http://mails.dpdk.org/archives/dev/2019-January/124242.html\n - Renamed non-blocking -> lock-free and NB -> LF to align with a similar\n change in the lock-free stack patchset:\n http://mails.dpdk.org/archives/dev/2019-March/125797.html\n - Added support for 32-bit architectures by using the full 32b of the\n modification counter and requiring LF rings on these architectures to be at\n least 1024 entries.\n - Updated to the latest rte_atomic128_cmp_exchange() interface.\n - Added ring start marker to struct rte_ring\n\nv4:\n - Split out nb_enqueue and nb_dequeue functions in generic and C11 versions,\n with the necessary memory ordering behavior for weakly consistent machines.\n - Convert size_t variables (from v2) to uint64_t and no-longer-applicable\n comment about variably-sized ring indexes.\n - Fix bug in nb_enqueue_mp that the breaks the non-blocking guarantee.\n - Split the ring_ptr cast into two lines.\n - Change the dependent patchset from the non-blocking stack patch series\n to one only containing the 128b CAS commit\n\nv3:\n - Avoid the ABI break by putting 64-bit head and tail values in the same\n cacheline as struct rte_ring's prod and cons members.\n - Don't attempt to compile rte_atomic128_cmpset without\n ALLOW_EXPERIMENTAL_API, as this would break a large number of libraries.\n - Add a helpful warning to __rte_ring_do_nb_enqueue_mp() in case someone tries\n to use RING_F_NB without the ALLOW_EXPERIMENTAL_API flag.\n - Update the ring mempool to use experimental APIs\n - Clarify that RINB_F_NB is only limited to x86_64 currently; e.g. ARMv8 has the\n ISA support for 128-bit CAS to eventually support it.\n\nv2:\n - Merge separate docs commit into patch #5\n - Convert uintptr_t to size_t\n - Add a compile-time check for the size of size_t\n - Fix a space-after-typecast issue\n - Fix an unnecessary-parentheses checkpatch warning\n - Bump librte_ring's library version\n\nGage Eads (6):\n ring: add a pointer-width headtail structure\n ring: add a ring start marker\n ring: add a lock-free implementation\n test_ring: add lock-free ring autotest\n test_ring_perf: add lock-free ring perf test\n mempool/ring: add lock-free ring handlers\n\n app/test/test_ring.c | 61 +--\n app/test/test_ring_perf.c | 19 +-\n doc/guides/prog_guide/env_abstraction_layer.rst | 10 +\n drivers/mempool/ring/Makefile | 1 +\n drivers/mempool/ring/meson.build | 2 +\n drivers/mempool/ring/rte_mempool_ring.c | 58 ++-\n lib/librte_ring/rte_ring.c | 92 ++++-\n lib/librte_ring/rte_ring.h | 334 ++++++++++++++--\n lib/librte_ring/rte_ring_c11_mem.h | 501 ++++++++++++++++++++++++\n lib/librte_ring/rte_ring_generic.h | 487 ++++++++++++++++++++++-\n lib/librte_ring/rte_ring_version.map | 7 +\n 11 files changed, 1494 insertions(+), 78 deletions(-)" }